Azure Vision in Foundry Tools

Clear guidance on how to use computer vision responsibly aligned with Microsoft responsible AI principles.

Create intelligent applications that recognize and verify human identity using facial recognition and liveness detection capabilities.

Intelligently crop images with automatic composition optimization.

Integrate vision capabilities through REST API or client libraries with no machine learning expertise required.

Fine-tune models for custom scenarios with few-shot learning capabilities. Next generation of Custom Vision with improved accuracy. Can start with as little as one image per label.

Microsoft automatically deletes images and videos after processing and does not train on customer data to enhance models.

Video data does not leave premises and is not stored on the edge where the container runs for privacy protection.

Uses pseudonymous identifiers to track bounding boxes without identifying individuals or detecting demographic information.

Built-in security with 34,000 full-time equivalent engineers dedicated to security initiatives, 15,000 partners with specialized security expertise, and over 100 compliance certifications including 50+ specific to global regions.

Extract printed and handwritten text from images with mixed languages and writing styles. Supports varied languages and writing styles.

Stream and analyze video content in real time with spatial analysis and object detection capabilities.

Track movement and analyze environments in real time using computer vision. Detects and locates human presence in video footage with bounding boxes, tracks movements, and outputs event data including zone entry/exit and directional line crossing.

Analyze visual content that pulls from more than 10,000 concepts and objects to detect, classify, caption, and generate insights. Includes automatic image captioning with natural language, smart crop, and image classification.

Automatically tag and classify images based on detected content.

Detect and identify objects within images and video streams for real-time analysis and understanding.